Passive Surface-Wave Imaging And Waveform Inversion In Urban Areas

China’s urbanization process is reaching its highest level in history.Advances in geophysical theory and technology can promote reforms in the quality,efficiency and dynamics of urban underground resources and space explorations.They will bring transformations from extensive expansion to data-driven scientific planning;from low cost-effective borehole explorations to scanning imaging under the constraints of multiple parameters;from surface building disaster prevention to fine utilization of underground space.Taking advantage of the abundant cultural noise in cities and towns,passive surface-wave imaging is well developed,eco-friendly,and easy to apply.However,in terms of the basic theory of seismic interferometry,especially for the(accuracy and resolution)performance of near-surface velocity imaging,there is still much room for research and improvement.Focusing on the high-resolution imaging of urban near-surface resources,this study applies the dense array technology to the geothermal field of Jinhua City,Zhejiang Province.Based on the collected data,a new idea of virtual source-receiver subnet extraction is tested.Compared with the conventional full-station network,the quantitative resolving performance is basically the same,and the theoretical processing efficiency of ray path data is nearly doubled.The tomographic phase velocity slices can be well matched with geological units and geographic features,and the inverted bodywave velocity model corresponds well with the borehole data.The shear wave velocity model describes the three-dimensional groundwater flow pipeline structure.The velocity profile accurately depicts the fractures and depression structures developed1 km below the drilling hole.Combined with temperature and resistivity logging information,comprehensive geological interpretations for groundwater migration and occurrence in this area are proposed to delineate the shallow geothermy to be the karst fissure water heated at a certain earth temperature gradient.Reliable dispersion measurement between two seismic stations is an essential basis of surface wave imaging.In classical passive seismic interferometry,the timefrequency information of the surface-wave signal depends on the phase of Green’s function within a certain frequency band.And the latter is usually obtained by the approximation of the cross-correlation function between station-pair noise records.Therefore,the cross-correlation function can be regarded as the logical beginning of the entire ambient noise imaging,and its formation is essentially controlled by the medium structure and the noise source distribution.For a near-surface seismic observation system with a finite time scale,the velocity structure of the medium is generally considered to be constant,and the spatiotemporal characteristics of the noise source can be represented by an overall model of the intensity distribution on a twodimensional plane.In the complex urban environment,the source model can rarely satisfy the(randomly)homogeneous distribution,and the ideal state of wavefield equipartitioning is hardly achieved during a short-period observation.The above problem is generally called the source azimuthal effect.It interferes with the crosscorrelation waveforms reflecting the real velocity structure to varying degrees,resulting in that the principle of the conventional Green’s function approximation are no longer strictly established,which will cause deviations in surface-wave dispersions.Focusing on the apparent velocity corrections of surface waves under the source azimuthal effects,this dissertation firstly demonstrates that under two special conditions of randomly homogeneous and inline source distributions,different source phase terms shall be adopted for Green’s function retrievals.After summarizing and classifying the previous methods,this dissertation further develops the ambient noise waveform inversion theory in view of the conflicting assumptions caused by the asynchrony between the noise source and the speed iteration.And this dissertation proposes to abandon the principle of Green’s function extraction for dispersion imaging,but to jointly invert source and velocity structures by fitting the cross-correlation waveforms.Waveforms intrinsically contain the features of travel-time,energy and asymmetry in cross-correlation functions.They come from the convolution of the noise source and the velocity structure,and can also be mapped to these two parameters through inversion,providing feasibility for joint inversion theory.Based on the average path velocity model between station pairs,the two Fréchet derivatives including the source misfit kernel and the velocity gradient are combined.And these two parameters in the model space are simultaneously inverted to iteratively achieve decoupling and optimization.Numerical tests reveal the mechanism and influence of the source distribution and the velocity model in waveform inversions,and show the necessity of joint inversion and the effectiveness of the algorithm.This research alleviates the longstanding theoretical and technical dilemma of velocity deviations caused by source azimuthal effects.It can provide high-quality time-frequency data with stronger physical meaning,and improve the reliability of next tomography.